in fact the decoder should _never_ panic. included here are tests that
cover the entire 32-bit instruction space and ensure that decoding and
display do not panic. these tests run uncomfortably slowly (1168s to
decode the 4b "instruction" sequences on my desktop), but verify that
panics are no longer an issue.

mostly confusion of pre/post-increment, operand widths, immediate
widths, things of that nature

16-bit instructions only, for now

Operands was an enum covering all forms of operands and data for each
operand for all instructions. this is hard to iterate, and hard to work
with when a single operand is the one of interest, so it's now replaced
by an array of `Operand` enum instances like other architectures.

in the course of this change, several forms of decoding are broken,
while adding support for some earlier-unsupported multiplies and
  instructions like msr and clz

also clearly note which document the comments mentioning page
numbers/figures is referencing